EP2759813B1 - Method and sensor for measuring the fill level of layered media - Google Patents

Description

The invention relates to a method for level measurement in a container having a first medium and at least one second medium arranged above, in particular a foam layer, according to the preamble of claim 1 and a trained for this method sensor.

A known method for level measurement is based on sending an electromagnetic signal into the container with the level to be measured and to evaluate the reflected signal. One way is to radiate the signal freely, as happens with the radar. Because of the uncontrolled wave propagation, the Time Domain Reflectometry (TDR) method is often preferred. It is based on the determination of transit times of an electromagnetic signal to determine the distance of a discontinuity of the line resistance. The difference to the radar is that the electromagnetic waves are not emitted to the outside, but are guided along a conductor. The conductor is designed as a monosonde or coaxial probe which is inserted vertically or obliquely into the tank and extends as close as possible to the bottom in order to cover the full measuring range.

In a TDR measurement, a very short electrical transmit pulse is fed into the conductor and passes through it towards the opposite end. If the impulse hits an impurity, which is equivalent to a change in the local characteristic impedance, part of the transmission energy is reflected back to the line input. From the transit time between the transmission of the transmitted pulse and the reception of the reflection, the position of the defect can be calculated in a location-specific manner. An important example of an impurity is an interface that separates two spatial regions with different physical or chemical properties, such as an interface between two media.

In order to be able to determine the time of reception precisely, the course of the received signal is sampled and fed to a digital evaluation. In this case, for example, local extreme points are searched in the waveform and their temporal location associated with a reflection at an interface.

Difficulties arise when multiple interfaces are present, as is the case with multiple media in a container, for example. It can be water that accumulates at the bottom of an oil tank. An important case is foaming on the surface of a liquid. Here you would often like to determine the level of the actual medium. However, a reflection at the interface with the foam can be confused with the measuring pulse of the medium or even merge with this measuring pulse in such a way that no sensible measurement result is delivered any more. A conventional evaluation algorithm, which is designed only for the detection of individual reflection signals, so can not handle such measurement situations.

Even an extension of a conventional evaluation algorithm by an Mehrechoauflösung of multiple boundary layers is thus often not given in practice condition in the way that the separation layers must have a certain distance, so that the resulting from the separation layers pulses in the echo signal have a sufficient distance from each other. It is even more difficult if the individual layers of the media are not even homogeneous. For example, in the case of foam, the density and thus the relative dielectric constant increase according to an unknown, usually monotonous function, so that numerous small overlapping pulses result in the echo signal.

A possible way out might be to increase the bandwidth of the transmit pulse. As a result, echo pulses that result from boundary layers located close to one another are rather separated. At the same time, however, the demands on the electronics design, such as the sampling rate and the subsequent signal acquisition, are also becoming more stringent. In addition, at very high frequencies beyond 10 GHz, the attenuation in a monoprobe by the skin effect increases sharply, so that the signal-to-noise ratio can be insufficient. In any case, an increase in bandwidth does not yet solve the problem of distinguishing the echoes of foam and medium, thus ensuring that the foam level is not output as the level of the medium, a falsified level of the medium or, in the worst case, no measured value.

Another way out might be to use a coaxial probe instead of a monaural probe. This can improve the resistance to foam because the echoes are mainly caused by variations in the field space within the coaxial tube. By an advantageous embodiment of the coaxial tube with only small openings, it is possible to minimize the ingress of foam into the tube. However, the possible uses are limited as compared to a monoprobe, since portions of the medium to be measured can be deposited in the coaxial tube. This can affect the availability of the sensor. In applications with hygiene requirements, the use is not possible. In addition, the level of the foam itself can not be determined in this way.

The approach of reducing foam formation from the outset by adding chemical agents or operating the plant with optimized process parameters is not feasible in most applications. Chemical agents affect the process medium and are usually unthinkable in the food industry anyway. Through optimized process parameters, the problem can be mitigated, especially in the operating phase when starting up foaming can rarely be sufficiently suppressed in this way.

The EP 2 365 302 A1 describes a method in which the distance to an interface and the relative dielectric constant of the interface-generating medium are detected a first time from a transition pulse at the transition of the probe into the container and the echo at the interface and a second time from an artifact pulse from the end of the probe is determined. If these measurements are not consistent with each other, the presence of a further boundary layer is concluded and the method is applied iteratively to the further boundary layer until all fill levels have been measured. This requires accurate information of the two reference pulses, ie the transition pulse from the beginning and the artifact pulse from the end of the probe. In order to achieve sufficient accuracy here, high demands are placed on the electronics, which lead to corresponding production costs. Even so, from the transition pulse of a monoprobe, which is exposed to a lot of other than a coaxial probe very much interference, do not close sufficiently accurate to the transmitted portion of the transmission energy. In addition, the artifact pulse at the end of the probe is delayed and attenuated by the propagation in the different media in a manner that is initially completely undefined and therefore is not available as a reliable reference in many measurement situations. Also, at low levels, too the artifact pulse of measuring pulses to be superimposed, so that the reference is even distorted.

The DE 100 51 151 A1 Figure 11 shows a TDR method for determining the positions of upper and lower interfaces of a first liquid floating on a second liquid of a container. However, it is not explained in more detail how the corresponding echoes are recognized and distinguished at the two interfaces. Furthermore, the evaluation is based on the artifact pulse, which is used in the DE 100 51 151 A1 is referred to as a reference feedback pulse. This has the above-described significant disadvantages.

From the US 6,724,197 B2 a level sensor is known with which the level of the lower of two stacked media can be determined. But this requires a special and expensive probe shape.

The US 5,723,979 A discloses a TDR level sensor for measurement in liquid mixtures. Only a probe shape is described. An evaluation method, certainly a determination of the distance to several boundary layers, is not described.

In the US Pat. No. 6,445,192 B1 For example, twin pulses at double interfaces are separated by a TDR sensor based on the slope behavior of the waveform. However, it is not explained how to measure the level of a medium under an interfering layer.

From the DE 101 96 640 T1 there is known an improved threshold setting for a radar level transmitter used using a TDR method to measure levels of multiple materials in storage containers. In doing so, multiple thresholds, each related to a specific material interface, are used to detect the levels of the various materials.

In the DE 100 44 888 A1 The dielectric constants and layer thicknesses of superimposed products are determined from the amplitude ratio of an echo curve of a pulse radar.

The post-published DE 10 2012 101 725 identifies in a signal curve of a TDR level sensor based on amplitude expectation values a first measuring pulse and a second measuring pulse, which arise at an interface with the medium to be measured or an interference or foam layer located above it. The expected values are in particular calculated from a known relative dielectric constant of the medium and a reference amplitude of an artefact pulse arising at the probe end when the container is empty.

The DE 10 2005 063 079 A1 discloses a method for detecting and monitoring a level of a medium in a container. In this case, a microwave signal is radiated into the container and evaluated the resulting echo function. The method is intended and suitable for processing an uneven surface of the medium, for example by means of bulk material formation, where a plurality of wanted echo signals are produced in the echo function due to the unevennesses. A group area of the useful echo signals is then identified and the measuring position determined as the centroid of the group area. This should represent a mean filling level.

In the DE 102 49 544 A1 Another method for level measurement according to the transit time principle is described. In order to take into account overlays of a first echo pulse from the contents with a second echo pulse from an interferer, run times are determined to a point X on the rising edge, to the maximum M of the overlay and to a point Y on the falling edge. The transit time to the product surface is derived from these terms in various embodiments in various ways. One embodiment provides a form factor for correcting overlay asymmetries, which compares the integral of the echo curve from the point X to the maximum M with the integral from the maximum M to the point Y.

Insofar as in the prior art it is even possible to dissolve several boundary layers, all these methods are based on detecting pulses at the boundary layers. At thin layers and especially foam, which can be formed in very different ways and can produce a variety of interlaced pulses, but fails the detection of individual pulses. The result is large measurement errors or, if in the recorded signal, a plurality of pulses is fused to each other beyond recognition and therefore no measuring pulse detected will, even error conditions of the sensor. TDR sensors are therefore traditionally not suitable for foam measurement.

It is therefore an object of the invention to improve the level measurement in a container with multiple layers.

This object is achieved by a method for level measurement in a container having a first medium and at least one second medium arranged thereon according to claim 1 and a sensor designed according to this method according to claim 15. The invention is based on the basic idea of not only evaluating the signal profile of a signal reflected in the container at the point where a border crossing is located, but also integrating the signal course from a reference position of the probe. This results in an integrated signal curve, which itself is a function of that probe position up to which each was integrated, or a function of the time monotonically connected with the probe position. As it were, the history of the signal components reflected before the border crossing is taken into account.

The waveform is preferably a discretely sampled time series of amplitude values of the reception intensity of the transmission signal reflected in the container, for example a microwave signal. Under integration here is forming a summarized measure of the waveform to a point of view understood. For this purpose, in the simplest case, the discrete samples are added up. This corresponds to an approximation of the integral by a step function. However, more precise integration methods, such as the trapezoidal method or the Runge-Kutta method, are also conceivable.

The invention has the advantage that a measurement up to the surface of a certain medium is also made possible if further media or interference layers such as foam are lying over it. By integrating, this is completely independent of the actual waveform and in particular of how well pulses can be identified in the waveform. Likewise, individual amplitude values of the waveform are not compared to thresholds or the like where there would always be large dependencies on hard-to-control constraints. As a result, the measurement becomes extremely robust and evaluates even signal waveforms meaningful in which no pulses are recognizable. Layers above the medium to be measured may even be inherently inhomogeneous, as in some cases of foam, which practically forms a multiplicity of boundary transitions between foam layers of different density. In contrast to the prior art, the invention makes it possible to practically use TDR sensors for foam measurement at all.

The reference position is preferably a probe start. This is not necessarily the physical probe start, for example, when a probe is bolted inside the sensor head, but at the beginning of the desired fill level measurement range. In the course of the signal, this reference position can be detected, for example, on a pulse that arises almost inevitably in the transition from sensor head to probe. However, the reference position can also be derived from structural properties of the sensor or calibrated ex works or at the site.

The signal transit time is preferably summed from signal travel time intervals which take into account a delay of the signal propagation at the probe position belonging to a respective signal delay time interval by a correction with the signal waveform integrated up to this respective probe position. The transmission signal penetrating into the second medium is delayed by the higher relative dielectric constant there compared with air. This should be in the conversion of the signal delay be compensated in an equivalent signal path for the determination of a distance of the border crossing and thus the level. However, the dielectric constant is regularly unknown or, in the case of inhomogeneous foam, it changes in extreme cases from sample point to sample point of the signal curve. Therefore, in this preferred embodiment, a time adjustment is made. The dielectric constant acting on the respectively considered probe position is estimated with the aid of the previously integrated signal curve in order to take into account the signal components reflected up to this probe position.

In the correction preferably additionally an environmental constant dependent on the specific wave propagation in the container. The specific environment of the probe, which is essentially determined by the container, especially its geometry, likewise influences the signal course. This applies more to a single-probe. In order to become independent of these influences in the estimation of the relative dielectric constant and the time adjustment, an environmental constant is determined or established. The time adjustment thus takes into account at least two influencing variables, namely the integrated signal curve and the ambient constant.

The environmental constant is preferably determined in a calibration measurement in the case of an empty container, in which an integral of the signal curve is formed over a region of an artifact pulse arising at the probe end. At the end of the probe, the signal energy remaining after reflections at the various border crossings is reflected. When the container is empty, this is practically the entire signal, resulting in a pronounced pulse. The pulse is well manageable or automatically identifiable and, moreover, is located at a fixed time position determined by the probe length, since there are no signal delays due to air deviating relative dielectric constants when the container is empty. By referring to the artifact pulse, numerous container and device dependent factors and thus the associated adjustment steps and calculations can be omitted. The artifact pulse measurement problems described in the discussion of the prior art do not arise here, as measured in a controlled, very simple situation with an empty container. Although the artifact pulse represents a particularly simple and reliable measure of the environmental constant, it is also possible in principle to model or measure the instrument and container-specific influences that have been compensated for become. It is also conceivable to use the transmit pulse or a transition pulse at the transition between sensor head and probe instead of the artifact pulse.

The border crossing is preferably detected by a comparison of the integrated signal curve with a threshold value. Thus, the signal progression from a time corresponding to the reference position is successively integrated up to an always considered time until the integrated signal waveform exceeds the threshold. The most recently considered time is considered as belonging to the position of the wanted border crossing and converted into a level. In this way, a very simple method is given to determine the border crossing stable and completely independent of the shape of the waveform, ie in particular regardless of whether can be identified in the waveform pulses.

For the comparison, the signal curve is preferably integrated a second time, but the second integration is based instead of the signal curve itself on its absolute value or the pointwise squared signal curve. The signal curve can also assume negative values. This happens, for example, when denser foam is above loose foam. Another example is an air bubble in a coaxial probe. For the determination of the signal propagation time, in particular the time adjustment, it is correct to consider this sign. Negative components in the signal curve also correspond to reflected transmission energy. Since the threshold measures how much transmission energy has already been lost due to reflections above a considered probe position, the sign should be ignored. This is done by taking the absolute value or squaring, the latter giving the physically correct measure of the energy. Other derived quantities instead of an absolute amount or pointwise squaring are conceivable.

The border crossing is preferably detected in an iterative method, which integrates the signal profile in steps predetermined by the sampling rate of the signal waveform from the reference position to a termination criterion, namely whether the signal waveform integrated up to the respective step or a corresponding integral of the absolute value of the signal waveform or of the Pointwise squared waveform exceeds the threshold. Such an iterative method is easily implementable in an algorithm for digital processing.

The signal propagation time to be determined is preferably initially set to zero and incremented in each step by an increment calculated as a quotient whose counter is the sum of the environmental constant and the integrated waveform and whose denominator is the difference between the environmental constant and the integrated waveform is formed. This is done in an iterative manner a time adjustment in order to compensate for the fact that time intervals in the signal curve do not directly correspond to signal propagation paths, but that a signal delay through a relative dielectric constant deviating from air occurs here within a medium. With the given calculation for the increment, this dielectric constant is estimated at each position. In this case, as already described above, the signal components already lost above the position by reflections, which are detected in the integrated signal course, and the specific environmental influences on the wave propagation, which are taken into account by the environmental constant and, in particular, an integral via the artifact pulse from the probe end.

The threshold value is preferably determined by a calibration measurement, in which only the first medium is in the container and a threshold integral of the signal profile is formed in the region of an echo signal at the border crossing to the first medium. This is a simple method to determine a suitable threshold under controlled conditions, namely without interfering second medium. This ensures that a pronounced, easily recognizable and measurable medium pulse is formed, via which the threshold integral is integrated. As always, Integral means a summary measure of the waveform in the considered range, to which the individual values of the waveform contribute. Such a measure can be determined, for example, by numerical integration, which can simply consist in summing up the values of the signal curve.

The threshold is preferably set to 90% to 110% of the threshold integral to detect the border crossing to the first medium. The threshold is thus almost or completely equated with the threshold integral. A certain relief from an identity makes sense, with a value below 100% ensuring that the threshold is indeed exceeded, even under the influence of noise, and thus a measured value is generated, which may underestimate the distance of the border crossing somewhat. Conversely, a value above 100% overestimates this distance and thus prevents a level value from being erroneously output within the upper, second medium.

The threshold value may preferably also be set to 5% to 30% of the threshold integral in order to detect the limit over all of the second medium. This threshold should therefore be close to zero and only be so large that not even noise is detected as a border crossing. Thus, the uppermost border crossing, ie the level of the second medium and in particular the foam surface is measured. The further the threshold deviates from zero, the more surely it differentiates from the noise, but increasingly also overestimates the distance of the uppermost border crossing, and thus measures increasingly deeper into the second medium.

It is also conceivable to set the threshold to both near 100% and slightly above zero, i. effectively use two thresholds. As a result, the uppermost border crossing to the second medium is measured and the border crossing to the first medium is measured. So you get both levels and thus the thickness of the layer of the second medium or the foam layer.

The border crossing to the second medium is preferably recognized by the fact that the integrated signal curve systematically differs from zero from the border crossing. As soon as the integrated waveform differs from zero, there was a first signal reflection, so at this point the signal has entered the uppermost medium. However, it is practically not enough just to compare with zero, as this is also due to noise effects. However, if one looks at the integrated signal curve over a larger range, it can be seen whether the deviation from zero is due systematically or only to the noise. This procedure is an alternative to a threshold value method, whereby both methods can be used and mutually plausibilized or the results can be offset against each other.

The waveform, the integrated waveform and / or the threshold are preferably scaled by a drift correction value. An important cause of drift is the temperature. The drift correction value is determined, for example, by comparing a transition pulse at the start of the probe or the transmit pulse itself with a previously learned reference transition pulse. This transition pulse is largely independent of the level, so that its changes can be attributed to drifts. It is then assumed that such drifts scale the transient pulse to the same extent as the rest of the waveform.

The signal curve is preferably rescaled with a level-dependent amplitude characteristic. In the upper measuring range, multiple reflections result, in which signal portions returning from a medium surface to the probe start are reflected again into the container. This can be repeated several times. In this way, the actual transmit pulse is transmitted directly and superimposed on these additional daughter pulses, which weaken further with each multiple reflection. The result is higher amplitudes in the upper measuring range. It is possible to model this effect or to measure it at different fill levels and in this way to determine a fill level-dependent amplitude characteristic which, as a function of a fill level, contains the signal increase due to multiple reflections as a factor. This amplitude characteristic will be higher than one in the upper measurement range and then drop to one at levels where these multiple reflections are no longer relevant, and even drop below unity at longer distances because of attenuation. By rescaling the waveform with the amplitude characteristic of this disturbing effect is compensated.

Fig. 1

eine schematische Querschnittsdarstellung eines Füllstandssensors in einem Behälter;

Fig. 2

ein schematisches Blockdiagramm eines Sensorkopfes des Füllstandssensors gemäß Figur 1;

Fig. 3

ein beispielhafter schematischer Signalverlauf bei gut getrennten Messpulsen von Grenzübergängen an zwei Medien als Vergleich;

Fig. 4

im oberen Teil einen Signalverlauf mit Kennzeichnung des als Grenzübergang zu einem ersten Medium erkannten Zeitpunkts in einem Beispiel eines Behälters mit einer Flüssigkeit und einer darüber angeordneten dicken Schaumschicht; im mittleren Teil eine Darstellung des aufintegrierten Signalverlaufs und des Schwellübertritts; im unteren Teil eine Darstellung der korrigierten Laufzeit im Vergleich mit einer Laufzeit in Luft;

Fig. 5

Darstellungen gemäß Figur 4 in einem Beispiel eines Behälters mit einer Flüssigkeit und einer darüber angeordneten dünneren Schaumschicht, deren Dichte kontinuierlich zunimmt;

Fig. 6

Darstellungen gemäß Figur 4 in einem Beispiel eines Behälters mit einer Flüssigkeit und einer darüber angeordneten dünneren Schaumschicht von anderer Konsistenz als in Figur 5; und

Fig. 7

Darstellungen gemäß Figur 4 in einem Beispiel eines Behälters, in dem sich nur Schaum befindet.

The invention will be explained below with regard to further advantages and features with reference to the accompanying drawings with reference to embodiments. The figures of the drawing show in:

Fig. 1

a schematic cross-sectional view of a level sensor in a container;

Fig. 2

a schematic block diagram of a sensor head of the level sensor according to FIG. 1 ;

Fig. 3

an exemplary schematic waveform with well separated measurement pulses from border crossings to two media as a comparison;

Fig. 4

in the upper part of a signal curve with marking of the detected as a border crossing to a first medium time in an example of a container with a liquid and a thick foam layer arranged above it; in the middle part, a representation of the integrated signal waveform and the threshold crossing; in the lower part a representation of the corrected running time in comparison with a running time in air;

Fig. 5

Representations according to FIG. 4 in an example of a container with a liquid and a thinner foam layer arranged above it, the density of which increases continuously;

Fig. 6

Representations according to FIG. 4 in an example of a container with a liquid and a thinner foam layer of a different consistency arranged above it than in FIG FIG. 5 ; and

Fig. 7

Representations according to FIG. 4 in an example of a container in which there is only foam.

FIG. 1 schematically shows a TDR sensor 10, which is mounted as a level sensor in a tank or container 12 with a liquid or generally a medium 14. Above the medium 14 is a second medium 16, for example foam in different consistency, or an oil layer as the second medium 16 on water as the actual medium 14. The relative permittivity or relative dielectric constant ε _{r 2} in the second medium corresponds neither to air nor to the medium 14 and may vary within the second medium 16 at different elevations.

The medium 14 forms a first boundary surface 18 or a first boundary junction with respect to the second medium 16, and the second medium 16 forms a second boundary surface 20 with respect to the air located above. The sensor 10 is designed to determine the distance to at least one of the interfaces 20, 22 and to derive from its known mounting position the level and if necessary based on the geometry of the container 12 and the amount of the medium 14 and the second medium 16. The basic procedure corresponds to the initially described and known TDR method, but differs in the still to be explained improved evaluation. Although the design of the sensor 10 as a liquid level sensor is a very important field of application, the sensor 10 can in principle also be used for other media. In particular, bulk or granules are to be considered.

The sensor 10 has a sensor head 24 with a controller 26, which is preferably accommodated on a common board. Alternatively, several circuit boards or flexprint carriers connected via plugs are conceivable. To the controller 24, a probe 28 is connected, which is designed here as an open monoprobe and so among other has the advantage that it can be easily cleaned for applications in the field of hygiene. The probe 28 may also be a coaxial probe.

The provided in the sensor head 24 controller 26 and its board is in FIG. 2 shown in a block diagram. The actual control and evaluation unit 30 is implemented on a digital building block, for example a microprocessor, ASIC, FPGA or a similar digital logic device as well as a combination of several such devices. As already described in the introduction, during a measurement, a signal, preferably a pulse, is applied to the probe 28 via a microwave transmitter 32 and the transit time of the reflex pulse generated at an interface 18, 20 and received in a microwave receiver 34 is measured by the distance of the interface 18, 20 and thus to determine the level in the container 12. The received signal of the microwave receiver 34 is digitized for evaluation after amplification in an amplifier 36 with a digital / analog converter 38.

FIG. 3 shows for illustrative purposes a schematic signal course with well separated measuring pulses 40, 42 at the interfaces 18, 20. In such a signal curve, conventional evaluations could easily locate the measuring pulses 40, 42. Difficulties arise when the measuring pulses 40, 42 run into each other. Until now, the problem of detecting measuring pulses 40, 42 becomes inextricable when the second medium 16 is inherently inhomogeneous and thus generates a multiplicity of overlapping pulses of unknown number.

The signal curve also contains further pulses, namely a transition pulse 44 at the beginning of the probe in the transition between the sensor head 24 and the probe 28 and the transmit pulse 46 itself. Further to the right and no longer shown, an artifact pulse is still generated at the end of the probe where, after the reflections at the interfaces 18, 20 remaining residual energy of the transmission signal is reflected.

In order to be able to determine fill levels of the media 14, 16 according to the invention even in the case of measurement pulses 40, 42 which are no longer identifiable, the approach described below is followed:

The amplitudes of the reflections in the signal path are a measure of the impedance change in the container 12 and thus also a measure of the change in the relative dielectric constants at a respective boundary layer 18, 20. The magnitude of a pulse in the signal curve depends on this change in the dielectric constant and the signal energy lost due to reflections already made above. The shape of the respective reflected pulse is retained, the pulses differ only in their height.

Thus, the signal curve S can be regarded as a linear superimposition of the pulse shape s ( t ) with only different scalings by the altitudes a _n and different temporal positions t _{n of} the n pulses: $$S={\displaystyle \underset{i=1}{\overset{n}{\Sigma }}}s\left(t-t_n\right)a_n\mathrm{,}$$

The constant pulse shape s ( t ) also means that an integral over the signal curve S is composed of summands scaled integrals over this pulse shape s ( t ): $$\int a_{n}s\left(\tau -t_n\right)\mathit{d\tau }=a_n\int s\left(\tau -t_n\right)\mathit{d\tau }\mathrm{,}$$

Both the signal curve S itself and its amplitudes, and thus also for the integral I ( t ) via the signal curve, are in each case functions of the dielectric constant ε _r . It also applies $$I\left(t\right)=\underset{0}{\overset{t}{\int }}S\left(\tau \right)\mathit{d\tau }=\underset{0}{\overset{t}{\int }}{\displaystyle \underset{i=1}{\overset{N}{\Sigma }}}s\left(t-t_n\right)a_n\mathit{d\tau }\mathrm{,}$$

Demnach ist der zu einem Zeitschritt ΔT der Signallaufzeit, beispielsweise einer Abtastperiode des A/D-Wandlers 38, äquivalente Weg bei Dielektrizitätskonstanten ε _r1, ε _r2 des ersten Mediums 14 beziehungsweise des zweiten Mediums 16

• für Luft: c ₀ΔT
• für das erste Medium 14: $\frac{{\mathrm{<figure-callout\; id="0"\; label="c"\; filenames="imgf0003.png,imgf0004.png"\; state="\{\{state\}\}">c</figure-callout>}}_0\mathrm{\Delta }T}{\sqrt{{\epsilon }_{\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="imgf0003.png,imgf0004.png"\; state="\{\{state\}\}">r</figure-callout>}1}}}$
  
  und
• für das zweite Medium 16: $\frac{{\mathrm{<figure-callout\; id="0"\; label="c"\; filenames="imgf0003.png,imgf0004.png"\; state="\{\{state\}\}">c</figure-callout>}}_0\mathrm{\Delta }T}{\sqrt{{\epsilon }_{\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="imgf0003.png,imgf0004.png"\; state="\{\{state\}\}">r</figure-callout>}2}}}\mathrm{.}$
  

As FIG. 3 illustrated, the signal propagation velocities c of the electric wave in the different layers differ, depending on the dielectric constant ε _r . While in air with practically ε _r = 1 the vacuum light velocity c ₀ can be assumed, within the media 14, 16 the velocity is delayed $\frac{1}{\sqrt{{\epsilon }_r}}\mathrm{,}$

Accordingly, the time step to a Δ T of the signal propagation time, for example, a sampling period of the A / D converter 38, equivalent way in dielectric constant is ε _{r 1, r 2} ε of the first medium 14 and second medium 16

• air: c ₀ Δ T
• for the first medium 14: $\frac{{\mathrm{<figure-callout\; id="0"\; label="c"\; filenames="imgf0003.png,imgf0004.png"\; state="\{\{state\}\}">c</figure-callout>}}_0\mathrm{\Delta }T}{\sqrt{{\epsilon }_{\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="imgf0003.png,imgf0004.png"\; state="\{\{state\}\}">r</figure-callout>}1}}}$
  
  and
• for the second medium 16: $\frac{{\mathrm{<figure-callout\; id="0"\; label="c"\; filenames="imgf0003.png,imgf0004.png"\; state="\{\{state\}\}">c</figure-callout>}}_0\mathrm{\Delta }T}{\sqrt{{\epsilon }_{\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="imgf0003.png,imgf0004.png"\; state="\{\{state\}\}">r</figure-callout>}2}}}\mathrm{,}$
  

This should preferably be taken into account in the evaluation by a corresponding time adjustment, because otherwise the distances of the interfaces 18, 20 are systematically overestimated. However, the dielectric constants are either lacking knowledge via the media 14, 16 or, for example, in the case of an inhomogeneous foam with density varying over the height, in principle not known in advance in advance.

If, as in a situation according to FIG. 3 If the amplitude A _{1 of} the measuring pulse 40 can be determined by the first medium 14, the associated dielectric constant ε _{r 1} can be calculated. In a procedure quoted in the introduction DE 10 2012 101 725 explained in more detail, to which reference is made for this purpose, one first determines a reference amplitude A _{end of} the artifact pulse from the _end of the probe. As a result, effects of the geometry of the container 12 and other signal changes can be taken into account. The reference amplitude A _end is for example calculated, simulated or determined by a repeated calibration measurement in the empty state of the container 12. It then applies $${\epsilon }_{\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="imgf0003.png,imgf0004.png"\; state="\{\{state\}\}">r</figure-callout>}1}=\frac{A_{\mathit{end}}-A_1}{A_{\mathit{end}}+A_1}\mathrm{,}$$

so dass insbesondere die Amplitude A_end des Artefaktpulses zu dessen Fläche proportional ist.Like all pulses, the artifact pulse also satisfies the relationship $$\int A_{\mathit{end}}{s}_{\mathit{end}}\left(t\right)\mathit{dt}=A_{\mathit{end}}\int s_{\mathit{end}}\left(t\right)\mathit{dt}.$$

so that in particular the amplitude A _{end of} the artifact pulse is proportional to its area.

A continuous transition of the wave impedance within the second medium 16 from the wave impedance of the air to the wave impedance of the first medium 14 results in a multiplicity of overlapping pulses. A single pulse evaluation as in FIG. 3 is then no longer possible, just as ε _{r 1} can not be calculated from A ₁ and A _end .

However, it is possible to evaluate integrals via the signal curve S and corresponding integrals via the artifact pulse or further integrals, which are determined as parameters in a controlled situation, for example when the container is empty or in the case of a container without a second medium 16. Thus, one no longer separates the signal curve S into discrete pulses, but assumes a hypothetical or actual reflection at each point because of a change in the dielectric constant. Then the location in the waveform S is located where as much signal energy has been reflected as in a comparison situation without the second medium 16, and this location is recognized as the first interface 18. In order to correctly determine their distance, the delay through the second medium 16 should preferably be taken into account when converting the signal propagation time into a path.

This procedure presupposes a simplifying assumption, although it is shown in practical measurements that, despite this simplification, good measurement results are obtained. It is namely assumed that the emitted signal energy arriving at the first interface 18 is independent of the course of the impedance changes above the first interface 18. It is assumed that the same energy loss of the signal by previous reflections, no matter in which way the dielectric constant of 1 in the upper air layer to ε _{r 1} at the first interface 18 changes, so if for example in a jump directly to the first interface 18, growing monotonically between the second interface 20 and the first interface 18 as with an increasingly dense foam or with any other behavior. The robust measurements against this approximation suggest that pathological processes are practically non-existent.

These general considerations can be illustrated by way of example in a concrete algorithm. However, the invention should not be limited to this algorithm, but also includes other implementations of the just described approach.

The first step will be to prepare in analogy to the DE 10 2012 101 725 measure the artifact pulse from the probe end when the container 12 is empty in order to detect the environmental influences of the specific installation situation of the TDR sensor 10. However, according to the integral method presented here, its amplitude is not determined, but the integral I _{end is formed} over the signal curve S in the region of the artifact pulse, but here, as stated above, there is a proportionality relationship.

In a second preparatory step, the container 12 is filled only with the first medium 14 and ensured that no foam as a second medium 16 is formed or this has dissolved. In this controlled environment is ensures that a clearly detectable measuring pulse 40 or medium pulse is generated at the first interface 18. About the measuring pulse 40 then the integral I _{medium is} formed. This measurement is carried out primarily to compensate for the different influences of the dielectric constant ε _{r 1 of} the first medium 14 and therefore could alternatively be replaced by a default or independent measurement of the dielectric constant ε _{r 1} , for example by subsequent calculation of I _medium on the basis of the known Pulse shape s ( t ).

gebildet, der eine Funktion des jeweils betrachteten Zeitpunkts t ist.During the actual measurement, the signal curve S is then integrated from an initial value t ₀ , for example corresponding to the start of the probe or a selected highest possible fill level, up to a respectively considered time t . So it becomes an integrated waveform $I_S\left(t\right)=\underset{0}{\overset{t}{\int }}S\left(\tau \right)\mathit{d\tau }$

formed, which is a function of the respective time t considered.

The position of the interface 18 is then found by comparing the integrated waveform I _S with the reference I _medium . For this purpose, it is determined, for example, for which t I _S ( t ) = K × I _medium , where K is a scaling factor. If K is chosen near unity, then t corresponds to the position of the first boundary surface 18. If K is selected only slightly above zero, the position of the second boundary surface 20 is found. A combination additionally makes it possible to measure the layer thickness of the second medium 17.

anhand des Terms $\frac{I_{\mathit{end}}-I_S}{I_{\mathit{end}}+I_S}$

erfolgen.The found t now corresponds to a signal transit time, which is composed of contributions of signal paths in air, in the second medium 16 and possibly also portions in the first medium 14. Therefore, levels are systematically underestimated unless the equivalent path is corrected by time adjustment based on the different wave propagation velocities. Such a time adaptation is possible using the integrated signal waveform I _S and becomes even more accurate if, in addition, the environmental influences are taken into account, for example, by I _end . Specifically, the time adjustment then in analogy to the above-cited calculation rule ${\epsilon }_{\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="imgf0003.png,imgf0004.png"\; state="\{\{state\}\}">r</figure-callout>}1}=\frac{A_{\mathit{end}}-A_1}{A_{\mathit{end}}+A_1}$

based on the term $\frac{I_{\mathit{end}}-I_S}{I_{\mathit{end}}+I_S}$

respectively.

In practical implementation, the signal curve S is typically present as a time series S [ i ] of discrete samples. Therefore, in a preferred embodiment, it is advisable to perform the evaluations iteratively and to calculate the integrated signal waveform I _S and the time adaptation step by step. For this purpose, initially all values are initialized to zero and then, starting from the beginning of the measuring range at t _0, iteratively adjusted step by step for the subsequent scans. The termination criterion is when I _{S reaches} a threshold K · I _medium derived from the reference value I _medium .

In Pseudocode an exemplary iterative evaluation can then be represented as follows:

vorzugsweise darauf geprüft werden, ob er im Intervall [0,1 ...1] liegt, da er sonst unplausibel wäre.It should be noted that I _end as the artifact pulse itself is negative. In addition, the correction value for the time adjustment should $\frac{I_{\mathit{end}}+I_S}{I_{\mathit{end}}-I_S}$

preferably be checked whether it is in the interval [0,1 ... 1], otherwise it would be implausible.

The result is the signal propagation time t , already corrected for the different signal propagation velocities in the media 14, 16, the units of which are sampling steps. The sampling rate is known, so that this can be directly converted into an absolute runtime and by means of the vacuum light speed c ₀ in a distance. Advantageously, the constant is calculated beforehand, which makes this conversion of signal propagation time in sampling steps in a distance by simple multiplication.

In particular, when the second medium 16 is foam, a special situation may still arise. It is conceivable that the dielectric constant does not increase monotonically with increasing distance, but also results in impedance jumps with the opposite sign. This happens, for example, when a denser foam layer rests on a thinner foam layer or air bubbles form. The result is reflected pulses with opposite signs or negative values in the signal curve S. As a result, I _S temporarily decreases despite the increasing penetration depth of the signal. Thus, the distance of the interface 18 is overestimated, in the worst case even the threshold K · I _medium never reached. However, the energy reflected back until the wave is in the medium 14 is approximately independent of the individual stages of impedance changes above it. Therefore, the termination condition should preferably be based on its absolute value or square instead of on the signed waveform S. For this purpose, a second integral I _{stop is} introduced, which sums up the absolute value or in a very analogous manner the square or a similar suitable moment.

This results in a further preferred embodiment of the evaluation, which can be summarized in this pseudocode:

In the pseudocode, the integration is carried out in the simplest way by adding up the values of the signal curve S. The invention also includes other methods of numerical integration. Similarly, the concrete formula for the Time adjustment interchangeable. In this case, the integrated signal course is preferably also used, because this takes into account the signal energy lost above the time point considered in the respective iteration step by reflection. However, the environmental influences can also be taken into account differently than via the artifact pulse by I _End , in particular by using a different pulse such as the transition pulse 44 or other information from the signal curve S.

In preparation, the waveform S can still be corrected in various ways. Thus, a rescaling to compensate drifts, in particular a temperature compensation, for example, by the transition pulse 44 at the start of the probe or the transmission pulse 46 is compared with a reference to estimate the drift influences.

Another way of compensating is to correct the waveform by an amplitude characteristic. Because the waveform S is excessive for small times corresponding to the upper portion of the container, because multiple reflections between the probe start and interface 18, 20 occur. This influence can be determined in advance and expressed in position- or time-dependent superelevation factors, which are then compensated by the amplitude characteristic. This too is in the DE 10 2012 101 725 explained in more detail, to which reference is made.

Finally, the invention according to the FIGS. 4 to 7 be illustrated as a medium 16 using the example of various foam layers.

FIG. 4 shows in the upper part of the waveform S with marking of the border crossing 18 detected to the first medium 14 time point, in one example of a container 12 with a liquid as the first medium 14 and an overlying thick foam layer as the second medium sixteenth

It should first be noted that in a first step, as described above, the higher curve has been corrected with an amplitude characteristic to compensate for an initial signal increase due to multiple reflections. The signal curve S which is actually the basis of the evaluation is therefore the lower curve, in which the gray area measured as integrated signal curve I _{S is also} highlighted is. At the vertical stroke approximately at the 70th scan, the first interface 18 was detected.

In the middle part shows the FIG. 4 the integrated to the respective sampling point waveform I _S , the gray area under the waveform S of the upper part of FIG. 4 gradually integrated. Approximately at the 70th scan corresponding to the position of the vertical bar in the upper part of the FIG. 4 is exceeded by the integrated waveform I _S a threshold shown as a thin gray line, whereupon the evaluation detects the first interface 18. This threshold is in several embodiments in K · I _medium . The dashed line corresponds to the total energy of the transmission pulse and can be exceeded at the earliest at the end of the probe

wächst die für die äquivalente Wegeausbreitung benötigte angepasste Zeit langsamer an, wie mit durchgezogener Linie gezeigt. Dem Abtastwert, der dem Abbruchkriterium entspricht, der in Figur 4 etwa beim 70sten Abtastwert liegt, kann die angepasste Signallaufzeit bis zu der ersten Grenzfläche 18 abgelesen werden und in einen Füllstand umgerechnet werden.Illustrated in the lower part FIG. 4 the time adjustment. The dashed line corresponds to the undisturbed wave propagation in air at maximum speed c ₀ . By the correction factor $\frac{I_{\mathit{end}}+I_S}{I_{\mathit{end}}-I_S}$

The adjusted time required for the equivalent path propagation increases more slowly, as shown by a solid line. The sample corresponding to the abort criterion, which in FIG. 4 is about the 70th sample, the adjusted signal propagation time can be read to the first interface 18 and converted into a level.

FIG. 5 shows quite another example of a container 12 with a liquid as the first medium 14 and a thinner foam layer arranged above it as the second medium 16, whose density increases continuously.

In the example of FIG. 6 the foam layer again has a different consistency and thickness.

FIG. 7 shows a case in which only foam is the only medium 14, 16 in the container 12. The threshold is exceeded here only very far back, depending on the setting of K only for values that would correspond to a negative level and therefore allow the conclusion that no medium 14 is present in the container 12. A conventional algorithm would have recognized the pronounced first pulse or one of the two daughter pulses and output the corresponding much too high level, despite the fact that the container was empty except for the foam.

If the fill level of the second medium 16 is to be determined, then this can be achieved by selecting a K close to zero according to the above. Alternatively, the second interface 20 is, however, also be seen that the integrated waveform I _S systematically different from zero. This is in each case in the middle part of the FIGS. 4 to 7 recognizable. At the same time, the adjusted time in the lower part of the FIGS. 4 to 7 For the first time systematically from the line of origin of the uncompensated time. With systematic deviation, it is meant here that a slight deviation is not yet to be evaluated on the basis of noise influences alone. This can be distinguished, for example, with the aid of a smoothing filter.

The invention has been described for a level measurement of a medium 14 with an overlying layer of a second medium 16. Of course, however, with the same procedure, a level of a medium is measured when there is no further layer over it. A generalization of the method to non-guided waves, ie a sensor 10 without a probe 28 such as a radar sensor, is also conceivable in principle.

Claims (14)

1. A method for filling level measurement in a container (12) having a first medium (14) and at least one second medium (16) arranged thereabove, in particular a foam layer, wherein an electromagnetic signal is transmitted, in particular along a probe (28) arranged in the container (12), and a signal curve (S) of the signal reflected in the container (12) is recorded; wherein a signal time of flight (t) up to a border transition (18, 20) to the first medium (14) or up to the second medium (16) is determined with reference to the signal curve (S ) and a filling level of the first medium (14) or a filling level of the second medium (16) is determined from the signal time of flight (t),
   characterized in that an integrated signal curve (I_s (t)), the integral starting from a reference position (t ₀), is formed as a function of a respective point in time (t), in that initially a reference value (I_Medium ) for the integrated signal curve (I_S (t)) is determined when the container (12) is filled with only the first medium (14), and in that the border transition (18, 20) is recognized by a comparison of the integral over the signal curve (I_S (t)) with the reference value (I_Medium ).
2. The method in accordance with claim 1,
   wherein the reference position (t ₀) is the probe start of a probe (28) arranged in the container (12).
3. The method in accordance with claim 1 or 2,
   wherein the signal time of flight (t) is summed up from signal time of flight intervals which take account of a delay in the signal propagation at the probe position belonging to a respective signal time of flight interval by a correction with the integral over the signal curve (I_S ) integrated up to this respective probe position, the probe position being the probe position of a probe (28) arranged in the container (12).
4. The method in accordance with claim 3,
   wherein an environmental constant dependent on the specific wave propagation in the container (12) additionally enters into the correction.
5. The method in accordance with claim 4,
   wherein the environmental constant is determined in a calibration measurement with an empty container (12), in which measurement an integral (I_end ) of the signal curve (S) is formed over a region of an artifact pulse arising at the probe end.
6. The method in accordance with claim 5,
   wherein the signal curve (S) is integrated for a second time for the comparison, but, instead of the signal curve (S) itself, its absolute amount (|S| ) or the signal curve (S ²) squared point-wise is used as the basis for the second integration (I_Stop ).
7. The method in accordance with claim 5,
   wherein the border transition (18, 20) is recognized in an iterative process which integrates the signal curve (S) in steps predefined by the sampling rate of the signal curve (S) from the reference position (t ₀) onward up to an abort criterion, namely whether the signal curve (I_S ) integrated up to the respective step (i) or a corresponding integral (I_Stop ) of the absolute amount (|S|) of the signal curve (S) or of the signal curve (S ²) squared point-wise exceeds the threshold value (K·I_Medium ).
8. The method in accordance with claim 7 and any of claims 3 or 4, wherein initially the signal time of flight (t) is set to zero and in every step (i) is increased by an increment $\left(\frac{I_{\mathit{end}}+I_S}{I_{\mathit{end}}-I_S}\right)$

   

   which is calculated as a quotient whose numerator is formed from the sum (I_end +I_S ) from the environmental constant ( I_end ) and the integral over the signal curve (I_S ) and whose denominator is formed from the difference (I_end -I_S ) from the environmental constant (I_end ) and the integral over the signal curve (I_S ).
9. The method in accordance with any of claims 5 to 8,
   wherein the reference value (I_Medium ) is formed as a threshold value integral of the signal curve (S)in a region of an echo signal (40) at the border transition (18) to the first medium (14).
10. The method in accordance with claim 9,
    wherein a threshold value (K·I_Medium ) is set to 90% to 110% of the threshold value integral (I_Medium ) to recognize the border transition (18) to the first medium (14) and/or wherein the threshold value (K·I_Medium ) is set to 5% to 30% of the threshold value integral (I_Medium ) to recognize the border transition (20) to the second medium (16).
11. The method in accordance with any of the preceding claims,
    wherein the border transition (20) to the second medium (16) is recognized in that the integral over the signal curve (I_S ) differs systematically from zero from the border transition (20) onward.
12. The method in accordance with any of the preceding claims,
    wherein the signal curve (S), the integral over the signal curve (I_s ) and/or the threshold value (K·I_Medium ) is scaled by a drift correction value.
13. The method in accordance with any of the preceding claims,
    wherein the signal curve (S)is rescaled by an amplitude characteristic dependent on the filling level.
14. A sensor (10), in particular a TDR filling level sensor, having a transmitter (32) and a receiver (34) for transmitting and receiving an electromagnetic signal, in particular a microwave signal, as well as having a control (30) which is configured to determine the filling level of a medium (14) and/or of a second medium (16) in a container (12) with reference to the signal time of flight, characterized in that the control (30) is configured to determine the filling level using a method according to any of the preceding claims.

Images